This is the title of the talk I gave on June 13th at the Psychology department of the University of Seville. The event was organised for the summer school of the department’s PhD program. I am grateful to professor David Saldaña for the kind invitation and for the interesting conversations.
In the talk, I discussed two types of causes for the observed low replicability rates in Psychology and other fields: a) Statistical and methodological, and b) systemic causes. Statistical/methodological causes include: a) inherent problems with the use of p-values as the only risk indicator for false positive results (ignoring the prior likelihood of the phenomena in question), and b) questionable research practices that tend to inflate the false positive rate.
Systemic causes refer to the ways academic research is currently validated, evaluated and disseminated. Conversations about these systemic issues tend to end up focusing on perverse system of academic incentives and the role of journals and evaluation committees in the preservation of this system.
I have presented proposed solutions for both types of problems, making emphasis on the importance of preregistration (to address methodological issues) and the transition to a scholarly communication model supported by a network of institutional repositories and overlay services offered by scientific societies, in line with the vision promoted by COAR and Open Scholar.